6omb AI

What Sets It Apart
Our platform introduced some standout features:

Voice AI Calling: We integrated with Bland.ai to introduce voice interactions. Users can now chat with their bots over voice calls, adding a human touch to digital conversations.

Custom Knowledge Base: Sales reps have the ability to upload their own knowledge bases. This means the chatbots can draw from a rich, representative-specific pool of information to answer queries.

Personalization: Every sales rep can tailor their chatbot to mirror their personality and style. It's about making digital conversations feel as natural as chatting with a friend.

Visitor Tracking: Our platform keeps tabs on web visitors and logs their chat history. Its like having a CRM built into your chatbot, making follow-ups and personalization easier.

Secure Access: We've implemented a robust authentication system, ensuring that only authorized sales reps can configure and access their chatbot settings.

Digging into the Tech
This project wasn't just about the end product; it was a journey through some pretty advanced tech:

Rapid Prototyping: Our first major win was developing a prototype that could seamlessly integrate a knowledge base with a Large Language Model (LLM), ensuring the chatbot could pull and process information effectively.

Front-End Development: Leveraging Vercel's NextJS framework, we crafted a user interface that's not only intuitive but also aesthetically pleasing, making chatbot interactions a breeze for users.

Tech Stack: We tapped into Vercel for hosting, which offered us the flexibility and scalability we needed. For the AI magic, we leaned on OpenAI's GPT-4 for its unparalleled natural language understanding and generation capabilities. The backend was powered by Postgres, with Langchain JS facilitating our interactions with the vector store, enriching our chatbot responses with contextually relevant information.

Cream of the Crop Features

AI-Driven Personalization: We harnessed some serious AI power to place users in fun, coffee-centric scenarios. Think of yourself sipping a latte on Mars!

Seamless Email Campaigns: Partnering with Beehiiv, we automated weekly emails to deliver these personalized images right to users' inboxes. Talk about a conversation starter!

Platform Integration: Users could link their image libraries through our secure APIs with the Conjure platform, deepening the personalization brew.

Cloud Flexibility: We built on Vercel and DigitalOcean to ensure CoffeeGPT could grow with its user base, handling more coffee lovers without breaking a sweat.

The Technical Grind

Cloud Infrastructure: We laid down a solid foundation with AWS, designing an architecture that's as robust as a double espresso.

API Magic: Our RESTful APIs made it easy to integrate with Beehiiv for email automation and Conjure for that personal touch.

Image Innovation: The project kicked off with us figuring out how to generate photorealistic images that put a smile on your face.

User Control: We made sure users could easily opt in or out, keeping them in the driver's seat of their CoffeeGPT journey.

Security First: Our APIs were locked down tighter than a coffee bean silo, with thorough testing to keep things running smoothly.

Tech Brewed Just Right

Cloud and Hosting: AWS for storage, Vercel for the front-end magic, and DigitalOcean for our Python-powered backend.

AI and Images: We mixed MTCNN for face detection with Fooocus and Juggernaut XL for those picture-perfect coffee scenarios.

Development Decaf: React and Next.JS for a sleek user interface, with Flask and Python brewing up the backend services.

Tech Docs: Crafting clear documents meant everyone knew the recipe for success.

Regular Refills: Updates and meetings kept us all on the same page, solving problems over a cup of virtual coffee.